those of the adult brain ( 27 ). On the other hand,
because there are changes in the relative pro-
portion of different cell types during devel-
opment ( 41 , 42 ), it is possible that in early
developmental stages, the main source of the
signal in Nissl-ST would be other cell types
(e.g., endothelial cells), thereby masking the
effect of the short glial rows. To demonstrate
the applicability of Nissl-ST to developmental
data, we compared the dataset of the adult
rhesus macaque (Fig. 4) with data from both
the embryonic stage at 120 days and the post-
natal stage at 14 days. These data suggest that
at 120 embryonic days, the density of glial cells
(or their progenitors) is very low compared
with that of the postnatal brain at 14 days
(fig. S17), with the latter having a much higher
similarity to the brain of an adult. Further-
more, at 120 embryonic days, the cells did not
seem to be organized in short rows. By con-
trast, at postnatal day 14, glial cells were already
organized in short rows along the axons’ex-
pected orientation. Therefore, future research
focusing on the intermediate stages of de-
velopment could shed light on the process
and ordering of myelination across white
matter pathways.
A common feature of Nissl-ST, PLI, and
structure tensor analysis on the basis of
myelin staining is their sensitivity to myelinated
(rather than nonmyelinated) axons. In addi-
tion, the proposed implementation of Nissl-
ST is two dimensional in nature. Hence, it is
unable to resolve the orientation of through-
plane fibers. Because glial cells associated with
out-of-plane axons cannot be easily identified
as such, they might introduce uncertainty
into the method’s orientation estimates. Future
studies directly comparing Nissl-ST and PLI in
the same slice could shed more light on their
differential sensitivity to both in-plane and
through-plane orientations. As an extension
of the proposed technique, the Nissl-ST meth-
od could be generalized to three-dimensional
datasets, such as the future high-resolution
version of BigBrain (which used silver stain-
ing for cell bodies rather than Nissl staining)
( 43 ). Notably, this would require specialized
tools for correcting staining variability between
sections ( 43 ). Such three-dimensional data-
sets could be integrated with neuroimaging
data using specially tailored toolboxes ( 44 ).
The glial framework—the patterned spa-
tial organization of glial cells in the white
matter—has been described in only a handful
of pathways, namely the rat fimbria ( 13 ), the
mouse corpus callosum ( 28 ), and the vervet
corpus callosum ( 29 ). Additionally, Pandya and
Schmahmann ( 3 ) briefly mentioned the glial
framework (i.e., the“glial matrix”) in the con-
text of the ILF and nearby tracts. It is a long-
standing question whether the glial framework
extends to all myelinated tracts, especially phy-
logenetically younger tracts. We found direct
evidence that this is the case for the corpus
callosum, the lenticular fasciculus, and the in-
ternal capsule in humans, rhesus macaques,
and vervet monkeys, as well as a U-fiber around
the occipitotemporal sulcus in the human brain.
The overall similarity of in-plane orientation
maps derived from Nissl-ST and from PLI
suggests that the structured organization of
glial cells along axons extends to other white
matter tracts.
The model-free approach of Nissl-ST reveals
a rich layer of information in Nissl-stained
brain slices; this information has been hitherto
unused. An advantage of Nissl-ST compared
with existing methods is the inherent co-
registration of the characteristic maps that
can be derived from white matter, as well as
the maps derived from cortical and subcortical
structures. Such coregistration would allow
for easy integration of Nissl-based atlases of
gray matter structures alongside maps of white
matter architecture. The abundance of Nissl-
stained resources in labs worldwide, as well
as the prevalence of digitized datasets and
open-source atlases including high-resolution
Nissl-stained slices ( 36 , 37 , 45 , 46 ), makes
the proposed technique readily applicable to
numerous datasets. Notably, such datasets
would allow for the comparison of fine-grained
features of fiber architecture across species
as well as between healthy and diseased brains,
and would provide a simple way to obtain a
histological reference for in vivo white mat-
ter mapping.REFERENCES AND NOTES- B. A. Wandell,Annu. Rev. Neurosci. 39 , 103 (2016).
- H. Takemura, M. Thiebaut de Schotten,Brain Struct. Funct.
225 , 1189 (2020). - J. D. Schmahmann, D. N. Pandya,Fiber Pathways of the Brain
(Oxford Univ. Press, 2006). - M. Axeret al.,Front. Neuroinform. 5 , 34 (2011).
- H. Takemuraet al.,eLife 9 , e55444 (2020).
- J. Menet al.,IEEE J. Sel. Top. Quantum Electron. 22 , 120
(2016). - M. Menzelet al.,Neuroimage 233 , 117952 (2021).
- H. R. Uedaet al.,Nat. Rev. Neurosci. 21 , 61 (2020).
- C. Stüberet al.,Neuroimage 93 , 95 (2014).
- A. Yendikiet al.,bioRxiv(2021.04.16.440223) [Preprint]
(2021). - K. Amunts, K. Zilles,Neuron 88 , 1086 (2015).
- N. Baumann, D. Pham-Dinh,Physiol. Rev. 81 , 871
(2001). - M. Suzuki, G. Raisman,Glia 6 , 222 (1992).
- J. Bigun, G. H. Granlund, J. Wiklund,IEEE Trans. Pattern Anal.
Mach. Intell. 13 , 775 (1991). - M. D. Budde, J. A. Frank,Neuroimage 63 , 1 (2012).
- W. Zhanget al., Structure Tensor Based Analysis of
Cells and Nuclei Organization in Tissues,IEEE Trans. Med.
Imaging 35 , 294 (2016). - J. Mollinket al.,Neuroimage 157 , 561 (2017).
- C. Baileyet al.,NMR Biomed. 32 , e4073 (2019).
- M. D. Budde, J. Annese,Front. Integr. Nuerosci. 7 ,3
(2013). - K. Amunts, A. Lindner, K. Zilles, The human brain project:
neuroscience perspectives and German contributions.
e-Neuroforum 5 , 43 (2014). - P. J. Basser, J. Mattiello, D. LeBihan,J. Magn. Reson. B. 103 ,
247 (1994). - J. D. Tournier, F. Calamante, D. G. Gadian, A. Connelly,
Neuroimage 23 , 1176 (2004).
23. A. Hornet al.,Neurology 92 , e1663 (2019).
24. E. J. L. Alhoet al.,Mov. Disord. 35 , 75 (2020).
25. M. M. Zeinehet al.,Cereb. Cortex 27 , 1779 (2017).
26. B. Jeurissen, A. Leemans, J.-D. Tournier, D. K. Jones, J. Sijbers,
Hum. Brain Mapp. 34 , 2747 (2013).
27. R. Reynolds, R. Hardy,J. Neurosci. Res. 47 , 455
(1997).
28. T. Tanakaet al.,Glia 69 , 2488 (2021).
29. M. Anderssonet al.,Proc. Natl. Acad. Sci. U.S.A. 117 , 33649
(2020).
30. J.-P. Lachaux, E. Rodriguez, J. Martinerie, F. J. Varela,Hum.
Brain Mapp. 8 , 194 (1999).
31. K. Schillinget al.,NMR Biomed. 30 , e3787 (2017).
32. J. Doucette, C. Kames, E. Hernández-Torres, A. Traboulsee,
A. Rauscher,bioRxiv(614008) [Preprint] (2019).
33. V. García-Marín, L. Blazquez-Llorca, J. R. Rodriguez,
J. Gonzalez-Soriano, J. DeFelipe,J. Comp. Neurol. 518 , 4740
(2010).
34. S. Mori, B. J. Crain, V. P. Chacko, P. C. M. van Zijl,Ann. Neurol.
45 , 265 (1999).
35. M. Guevara, P. Guevara, C. Román, J. F. Mangin,Neuroimage
212 , 116673 (2020).
36. J. K. Mai, M. Majtanik, G. Paxinos,Atlas of the Human Brain
(Academic Press, ed. 4, 2015).
37. J. A. Milleret al.,Nature 508 , 199 (2014).
38. J. D. Yeatman, B. A. Wandell, A. A. Mezer,Nat. Commun. 5 ,
4932 (2014).
39. J. Fitzsimmons, M. Kubicki, M. E. Shenton,Curr. Opin.
Psychiatry 26 , 172 (2013).
40. S. Hickman, S. Izzy, P. Sen, L. Morsett, J. El Khoury,Nat.
Neurosci. 21 , 1359 (2018).
41. R. R. Sturrock,J. Anat. 122 , 521 (1976).
42. R. K. Sigaard, M. Kjær, B. Pakkenberg,Cereb. Cortex 26 , 89
(2016).
43. K. Amuntset al.,Science 340 , 1472 (2013).
44. C. Paquolaet al.,bioRxiv(2021.05.04.442563) [Preprint]
(2021).
45. S. Mikula, I. Trotts, J. M. Stone, E. G. Jones,Neuroimage 35 ,9
(2007).
46. G. Striedter, The Zoomable Human Brain Atlas,
https://zoomablebrain.bio.uci.edu/.
47. R. Schurr, A. Mezer, Nissl-ST, Version 1.0, Zenodo (2021);
https://doi.org/10.5281/zenodo.5245155.
ACKNOWLEDGMENTS
We thank B. Wandell, A. Rokem, S. Berman, S. Filo, E. Drori,
H. Takemura, J. Yeatman, and K. Weiner for helpful comments on this
manuscript. Dataset 1 was provided by J. Mai and M. Majtanik.
Dataset 2 © 2010 Allen Institute for Brain Science. BrainSpan Atlas
of the Developing Human Brain is available from human.brain-
map.org. Dataset 3 (Macaca mulatta) was provided by E. G. Jones.
Dataset 4 (Chlorocebus aethiops) was provided by NeuroScience
Associates. Datasets 5 and 6 © 2010 Allen Institute for Brain
Science. NIH Blueprint Non-Human Primate (NHP) Atlas is
available from http://www.blueprintnhpatlas.org.Funding:This research
was supported by the Israel Science Foundation (grant 1169/20)
given to A.A.M., and by the Jerusalem Brain Community (JBC) award
granted to R.S.Author contributions:Conceptualization,
methodology, formal analysis, visualization, and writing: R.S. and
A.A.M. Supervision: A.A.M.Competing interests:The authors
declare no competing interests.Data and materials availability:
All original datasets used in this work can be accessed on the
websites listed in the methods section. The MATLAB code for
calculating and visualizing the structure tensor is available along
with the code for downloading example data on Zenodo ( 47 ).
Fig. S19 shows the expected outputs of the example code.SUPPLEMENTARY MATERIALS
science.org/doi/10.1126/science.abj7960
Materials and Methods
Supplementary Text
Figs. S1 to S19
Tables S1 and S2
References ( 48 Ð 71 )
MDAR Reproducibility Checklist4 June 2021; accepted 14 September 2021
Published online 7 October 2021
10.1126/science.abj7960SCIENCEscience.org 5 NOVEMBER 2021•VOL 374 ISSUE 6568 767
RESEARCH | REPORTS